Professional Documents
Culture Documents
O
ne of the changes usually described with a Rayleigh probability
This tutorial describes the that modern digi- density function. A Rayleigh, or fading, signal
principles of bit-error rate, tal communica- path is not “noise” in the intuitive sense of the
the factors that affect it, tions systems has familiar hissing sound of “white noise,” but it
and the methods for its brought to radio engi- is a random process that is analyzed in the
measurement, but with a neering is the need for same manner as Gaussian noise.
recommendation that end-to-end performance Without going into detail, the mathemati-
readers do further study of measurements. The mea- cal representations of these functions allow
its mathematical basis sure of that performance the analysis of a system to help predict its per-
is usually bit-error rate formance. Further study of noise and statisti-
(BER), which quantifies the reliability of the cal signal analysis is highly recommended.
entire radio system from “bits in” to “bits out,”
including the electronics, antennas and signal Quantization errors
path in between. Quantization errors also reduce BER per-
On the surface, BER is a simple concept— formance, through incorrect or ambiguous
its definition is simply: reconstruction of the digital waveform. This is
also a described by a probability function that
BER = Errors/Total Number of Bits defines the likelihood that a digital transition
or edge detection error will occur. These errors
With a strong signal and an unperturbed are primarily a function of the accuracy of the
signal path, this number so small as to be digital-to-analog and analog-to-digital conver-
insignificant. It becomes significant when we sion processes, and are related to the number
wish to maintain a sufficient signal-to-noise of bits used at these points in the circuit.
ratio in the presence of imperfect transmis- The accuracy of the analog modulation/
sion through electronic circuitry (amplifiers, demodulation process and the effects of filter-
filters, mixers, and digital/analog converters) ing on signal and noise bandwidth also affect
and the propagation medium (e.g. the radio quantization errors.
path or optical fiber).
Any in-depth analysis of the processes that Eb/No and BER
affect BER require significant mathematical BER can also be defined in terms of the
analysis. I will not attempt to do that in this probability of error (POE),
short introduction and overview.
1
POE = (1 − erf ) Eb / No
Noise and BER 2
Noise is the main enemy of BER perfor-
mance. Noise is a random process, defined in where erf is the error function, Eb is the ener-
terms of statistics. The noise introduced by the gy in one bit and No is the noise power spec-
circuitry is described with a Gaussian proba- tral density (noise power in a 1 Hz band-
bility density function, while the signal path is width). The error function is different for the